Astronomy 730. Milky Way

Transcription

1 Astronomy 730 Milky Way

2 Outline } The Milky Way } Star counts and stellar populations } Chemical cartography } Galactic center and bar } Galactic rotation } Departures from circular rotation

3 Modeling the Milky Way Galaxy (MWG) } What is the stellar distribution? } How big is the Milky Way? } Does (where does) the disk have an outer truncation? } Does it have a bar? } How do the stars move in the galaxy? } Galactic rotation and Oort s constants

4 Star Counts } Formalism } N(M,S)= (M,S)D(r)r 2 dr } N = # of field stars of a given absolute magnitude (M) and spectral type (S) in the galaxy } = stellar luminosity function (# pc -3 for some spectral type) } D(r) = density distribution may also depend on M,S: D(r,M,S) Alternatively, may also depend on r } The simplest thing we can actually observe: } A(m) = # of stars of some apparent magnitude, m. } A(m) = (M) D(r) Ω r 2 dr } Ω = solid angle of survey } If we have colors, we might get a crude A(m,S) } but without distances or some luminosity indicator, we can t break the dwarf-giant degeneracy (e.g., K V vs K III).

5 Star Counts: Infinite Euclidean Universe } A(m) represents the differential counts, i.e., number of stars per apparent magnitude interval } Knowing the geometry (locally Euclidean), the count slope (da/dm) tells us about the spatial distribution of sources. } For a uniform space-distribution of sources it is straightfoward to show } d(log A)/dm = 0.6m + constant } This leads to Olbers paradox: } l(m) = l 0 dexp(-0.4m) } L(m) = l(m)a(m) } L tot = l(m)a(m) dm = } The distribution of stars in the galaxy must be spatially finite } What about the universe of galaxies? log A(m) Euclidean slope = 0.6m m What would this imply?

6 The Malmquist Bias & the Night Sky } What stars do you see when you look up at the night-time sky? } Does this make sense given what you know about the HR diagram } Malmquist bias: } You can see brighter objects farther away to a fixed m } Volume increases as d 3 L 1,d 1 L 2,d 2 L 3,d 3 For a uniform space-density, the observer is biased toward finding intrinsically luminous objects.

7 The Malmquist Bias & the Night Sky } What stars do you see when you look up at the night-time sky? } Does this make sense given what you know about the HR diagram } Malmquist bias: } You can see brighter objects farther away to a fixed m } Volume increases as d 3 L 1,d 1 L 2,d 2 L 3,d 3 For a uniform space-density, the observer is biased toward finding intrinsically luminous objects.

8 Star Counts & The Malmquist Bias } What s the mean magnitude of stars with apparent magnitude, m? } M(m) = M (M)D(r)r 2 dr / (M)D(r)r 2 dr } Recall A(m) = (M) D(r)Ωr 2 dr } Assume the stellar luminosity function (LF) is Gaussian for a given spectral type: } (M,S) = 0 /(2π) 1/2 exp[-(m-m 0 ) 2 /2 2 ] } M 0 = mean magnitude } 0 = # pc -3 for some spectral type } is the distribution width Zheng et al. (2004, ApJ, ): MW disk M-dwarfs, I-band, HST

9 Star Counts & The Malmquist Bias } Push through the integral for M(m) and move things around a bit. } M(m) = M 0 [ 2 /A(m,S)] [da(m,s)/dm] } Or: M(m) M 0 = - 2 d lna / dm This is for the specific case of a Gaussian LF, but it can be generalized. } If there are more stars at faint magnitudes, then the stars at some m are more luminous than the average for all stars in a given volume } This will come back to bite us with a vengeance when considering distant galaxy counts A note on logarithmic derivatives: d ln x = dx /x This is nice because it normalizes the gradient to the amplitude of the signal, i.e., dimensionless. Therefore often used in Astronomy

10 Space Densities: Local Neighborhood Volume-limited Brightness-limited

11 Stellar Luminosity Function } Measure for a distance(volume)-limited sample } Bahcall & Soneira (1980) used: } (M) = [n * 10 (M-M * ) ] / [1+10 -( - ) (M-M * ) ] 1/ } n * = 4.03 x 10-3 } M * = 1.28 } α = 0.74, = 0.04, 1/ = 3.40 } See also Figure 2.4 in S&G. } Basic results } 10 5 times more G stars than O stars } Nearby stars tend to be } low-luminosity and apparently faint } Average M/L = 0.67 (M/L)

12 Star Counts: The Disk } Star counts (z direction) } n(z) proportional to exp(-z/z 0 (m)), } z 0 (m) is the scale height (and, yes, it does vary with magnitude) } More importantly, z 0 varies with spectral type: } young : old è small : large } WHY? } Reid & Majewski (1993) } Thin disk with z 0 = 325 pc Pop I } Thick disk with z 0 = 1200 pc Pop II } The Disk } I(R) = I(0)exp[-R/h R ] } tough to measure in our own galaxy, but measurable in other disks pretty easily. } Disk is really a double exponential: } I(R,z) = I(0,0)exp(-z/z 0 -R/h R ) young old Du et al A&A , stellar density perpendicular to the plane

13 Star Counts: The Halo } Halo stars } Halo stars are faint, need an easy to find tracer RR Lyrae stars } Stellar density falls off as r -3 } Looks like the distribution of globular clusters, which you also get from RR Lyrae stars } We will come back to this density fall-off when we consider the rotation curve.

14 Recall: RR Lyrae Stars } HB stars in the instability strip } Solar mass } Opacity driven pulsations yield variability which is correlated with M } M = -2.3±0.2 log(p) 0.88±0.06 (with some additional variation due to metallicity) } Old, low mass stars (hence good tracers of the halo) } Higher mass (farther up the instability strip you ll find Cepheids)

15 M33 LF } Outer fields of M33 } (Brooks et al. 2004, AJ, 128, 237) Counts proportional to luminosity function of all stars in M33 halo, corrected for contamination and completeness. TRGB at I = 20.7 gives distance modulus.

16 Galactic Center } We ll talk about the center again when we discuss AGNs } The optical view: } The VLA 1.4 Ghz view:

17 Galactic Center: Distance } Use RR Lyraes + other stellar tracers } Use the globular cluster population, OH/IR stars in the bulge } Get mean distances è 8.5 kpc } Proper motion studies of Sgr A* } Look for maser emission } Follow maser proper motion + observed velocity è distance (7.5 kpc)

18 Galactic Bar } Lots of other disk galaxies have a central bar (elongated structure). Does the Milky Way? } Photometry what does the stellar distribution in the center of the Galaxy look like? } Bar-like distribution: N = N 0 exp (-0.5r 2 ), where r 2 = (x 2 +y 2 )/R 2 + z 2 /z 0 2 } Observe A(m) as a function of Galactic coordinates (l,b) } Use N as an estimate of your source distribution: } counts A(m,l,b) appear bar-like } Sevenster (1990s) found overabundance of OH/IR stars in 1 st quadrant. Asymmetry is also seen in RR Lyrae distribution. } Gas kinematics: V c (r) = (4πG /3) 1/2 r è we should see a straight-line trend of V c (r) with r through the center (we don t). } Stellar kinematics again use a population of easily identifiable stars whose velocity you can measure (e.g. OH/IR stars). } Similar result to gas.

19 l=30 o l=-16 o (344 o ) 1 st quadrant 4 th quadrant 2 nd quadrant 3 rd quadrant Galactic Longitude Asymmetry Credits: Robert Hurt (SSC/JPL/NASA) Bob Benjamin (UWW)

20 Chemical Cartography } The APOGEE-1 view from SDSS-III abundance height metallicity radius Also: RAVE, SEGUE, GALAH/HERMES, APOGEE-2 Hayden+ 15

21 Galactic Rotation: A Simple Picture } Imagine two stars in the Galactic disk; the Sun at distance R 0, the other at a distance R from the center and a distance, d, from the Sun. The angle between the Galactic Center (GC) and the star is l, and the angle between the motion of the stars and the vector connecting the star and the Sun is. The Sun moves with velocity, V 0, and the other star moves with velocity, V. } See Figure 2.19 in S&G.

22 Relative motion of stars } Radial velocity of the star } V r = V cos V 0 sin l } now use law of sines to get } V r = ( * - 0 ) R 0 sin l, } ω is the angular velocity defined as V/R. } l is the Galactic longitude } Transverse velocity of the star } V T = ( * - 0 ) R 0 cos l * d R 0 Sun l R α d V 0 α V T GC

23 Longitudinal dependence } 90 o l 180 o } larger d } R > R 0 } * *< 0 } this means increasingly negative radial velocities } 180 o l 270 o } V R is positive and increases with d 90 o l 180 o 180 o l 270

24 Longitudinal dependence } 0 o l 90 o } starting with small R, large } At some point R = R 0 sin(l) and d = R 0 cos(l) } Here, V R is a maximum è tangent point. } We can derive * (R) and thus the Galactic Rotation Curve! } Breaks down at l < 20 o (why?) and l > 75 o (why?), but it s pretty good between 4-9 kpc from Galactic center.

25 Galactic Rotation Curve Best fit yields V C ~ 220±10 km/s, and its flat! Nakanishi & Sofue (2003, PASJ, 55,191)

26 Galactic rotation } Inner rotation curve from tangent point method è V circ, = 220 km s -1 } Derived from simple geometry based on a nearby star at distance, d, from us. } Tangent point where R = R 0 sin l and d = R 0 cos l : Observed V R is a maximum } Outer rotation curve from Cepheids, globular clusters, HII regions è anything you can get a real distance for } Best fit: (220±10 km/s) depends on R 0 (think back to the geometry) } Yields 0 = V 0 /R 0 = 29±1 km s -1 kpc -1 Sun l GC R 0 α d V 0 R α V T

27 Rotation model } Observations of local kinematics can constrain the global form of the Galactic rotation curve } Components of rotation model: } Oort s constants which constrain local rotation curve. } Measurement of R 0 } Global rotation curve shape (e.g., flat) } Oort s constants A and B: } 0 = V 0 /R 0 = A-B } (dv/dr) R0 = -(A+B) } V c, = R 0 (A-B)

28 Oort s Constant A: Disk Shear } Assume d is small } this is accurate enough for the solar neighborhood } Expand ( * - 0 ) = (d /dr) R0 (R-R 0 ) } Do some algebra. } V R = [(dv/dr) R0 (V 0 /R 0 )] (R-R 0 ) sin l } If d << R 0, } (R 0 -R) ~ d cos(l) } V R = A d sin(2l) } where A = ½[(V 0 /R 0 ) (dv/dr) R0 ] } This is the 1 st Oort constant, and it measures the shear (deviation from rigid rotation) in the Galactic disk. } In solid-body rotation A = 0 } If we know V R and d, then we know A and (d /dr) R0

29 Oort s Constant B: Local Vorticity } Do similar trick with the transverse velocity: } V T = d [A cos(2l)+b], and } l = [Acos(2l)+B]/4.74 = proper motion of nearby stars } B is a measure of angular-momentum gradient in disk (vorticity: tendency of objects to circulate around) } B = ± 0.6 km/s/kpc } A measure of angular-momentum gradient in disk } 0 = V 0 /R 0 = A-B } (dv/dr) R0 = -(A+B) } Observations of local kinematics can constrain the global form of the Galactic rotation curve

30 Measuring Oort s Constants } Requires measuring V R, V T, and d } V R and d are relatively easy } V T is hard because you need to measure proper motion } (arcsec yr -1 ) = V T (km s -1 )/d (pc) = V T /4.74d } Proper motions + parallaxes } A = km s -1 kpc -1, B = ± 0.6 km s -1 kpc -1 } The interesting thing you also measure is the relative solar motion with respect to the Local Standard of Rest (LSR)

31 Solar Motion } Stellar motion in the disk is basically circular with some modest variations. } There is an increase in the velocity dispersion of disk stars with color è age } Seen in vertical, radial, and azimuithal dimensions } Results in v correlation with (B-V) } What about the thickness of the disk? } Disk stars come in all different ages, but tend to be metal rich

32 Solar Motion } LSR velocity of something moving in a perfectly circular orbit at R 0 and always residing exactly in the mid-plane (z=0). } Define cylindrical coordinate system: } R (radial) } z (perpendicular to plane) } (azimuthal) } Residual motion from the LSR: } u = radial, v = azimuthal, w = perpendicular } Observed velocity of star w.r.t. Sun: } U * = u * -u, etc. for v, w } Define means: } <u * > = (1/N) u *, summing over i=1 to N stars, etc. for v,w } <U * > = (1/N) U *, etc for V,W

33 Solar Motion } Assumptions you can make } Overall stellar density isn t changing } there is no net flow in either u (radial) or w (perpendicular): } <u * > = <w * > = 0. } If you do detect a non-zero <U * > or <W * >, this is the reflection of the Sun s motion: } u = -<U * >, w = -<W * >, v = -<V * > + <v * > } Dehnen & Binney 1998 MNRAS (DB88) } Parallaxes, proper motions, etc for solar neighborhood (disk pop only) } u = ± 0.36 km s -1 (inward; DB88 call this U 0 ) } v = 5.25 ± 0.62 km s -1 (in the direction of rotation; DB88 call V 0 ) } w = 7.17 ± 0.38 km s -1 (upward; DB88 call this W 0 ) } No color dependency for u and w, but for v.

34 Solar Motion } Leading & Lagging } Stars on perfectly circular orbits with R=R 0 will have <V> = 0. } Stars on elliptical orbits with R>R 0 will have higher than expected velocities at R 0 and will lead the Sun } Stars on elliptical orbits with R<R 0 will have lower than expected velocities at R 0 and will lag the Sun } Clear variation in v with (B-V)! } Why? } Why only v and not u or w? } We can also measure the random velocity, S 2, and relate this to v. This correlation is actually predicted by theory (as we shall see)! } S = [<u 2 > + <v 2 > + <w 2 >] 1/2

35 Parenago s Discontinuity Clues to disk evolution: τ ms è mean age Hipparcos catalogue: geometric parallax and proper motions Binney et al. (2000, MNRAS, 318, 658) S = S 0 [1+(t/Gyr) 0.33 ] ç random grav. encounters S = 0 8 km s -1 ç why might this be? See also Wielen 1977, A&A, 60, 263

36 Parenago s Discontinuity: the disk } The disk is observed to be well described by a double exponential in radius (R) and vertical height (z) } Revisit nomenclature from lecture 6 to be consistent with S&G: } ρ(r,z) = ρ 0 exp(-z/h z ) exp(-r/h R ) young old } ρ is matter density, e.g., in stars ρ * = n * m * } Integrate ρ(r,z) in z to get Σ(R), e.g. M pc -2 } Σ(R) = ρ(r,z) dz } Multiply by the mass-to-light ratio (M/L = Υ) to get I(R), the surface-brightness : I(R) = Υ -1 Σ(R) } μ(r) often is used to denote surface-brightness in magnitudes arcsec -2. } μ(r,θ) would be surface-brightness at location R,θ in the disk (cyclindrical coordinates) } Integrate Σ(R) in R to get total mass within a given radius M(R), or I(R) to get total light } M(R) = 2π Σ(R) r dr } Why is the distribution exponential in radius? } This is hard to answer definitively, but it is an observed fact. } Why is the distribution exponential in height? } Here we will attempt to get a better physical standing in coming lectures. young old

37 The Halo: Clues to formation scenario? } Layden 1995 AJ } Age of halo RR Lyrae stars > 10 Gyr } -2.0 < [Fe/H] < -1.5 ; V rot / los ~ 0 ; los ~ km s -1 } -1.0 < [Fe/H] < 0 ; V rot / los ~ 4 ; los ~ 50 km s -1 } Relative to LSR } <U> = -13 km s -1 } <W> = -5 km s -1 } <V> [Fe/H]<-1.0 = 40 km s -1 } <V> [Fe/H]>-1.0 = 200 km s -1 Velocity dispersion defined: 2 los = (v los v)2 F(v los ) dv los or, los = ((v v)2) 1/2 where F(v los ) = velocity distribution function } Conclusion: there is an extended old, metal poor stellar halo dominated by random motions with very little, if any, net rotation (0 < V < 50 km/s)

38 Globular Cluster Population } Harris, W.E Star Clusters } ~150 globular clusters in MWG } Distribution is spherically symmetric, density falls off as R GC -3.5 } Bimodal metallicity distribution } [Fe/H] ~ -1.7 (metal-poor) è found in halo } [Fe/H] ~ -0.2 (metal rich) è found in bulge

39 Measuring Galactic Rotation } Gas: } Good because the MW is optically thin at CO (mm) and HI (21cm) wavelengths } Bad because you have to use the tangent method } essentially impossible to measure distances } Stars: } Good because you can measure distances directly } Bad because it is difficult to measure distances for distant or faint stars } Bad because traditional studies are done in optical, which can t penetrate mid-plane dust Sun l R 0 α d V 0 R α V T } enter the Sloan Digital Sky Survey (SDSS): GC

40 Measuring Galactic Rotation: Example } Select stars of a single spectral type. A stars Xue et al. 2008

41 Measuring Galactic Rotation: Example } Distinguish between blue horizontal branch stars and blue stragglers (MS) so the luminosity is known } Infer distances

42 Sight Lines

43 Measuring Galactic Rotation: Example } Determine the spatial distribution w.r.t. the GC è } Measure the observed distribution of line-of-sight velocities (ê V los ), and the dispersion of these velocities, σ los, as a function of Galactic radius î

44 Measuring Galactic Rotation: Example } And now the trick: Estimate circular velocity (the rotation curve) from the velocity-dispersion data. } So we need to learn some dynamics

45 Why Dynamics? } We can then also interpret the data in terms of a physical model: Mass decomposition of the rotation curve into bulge, disk and halo components : è Dark Matter è Stellar M/L Υ * è The IMF è Missing physics